What’s the best way to build a rocket?

If you’re Tim Ellis, the CEO of the space launch services start-up Relativity Space, the answer is obvious: You print it.

Ellis quietly formed Relativity Space in 2015 with Jordan Noone, after each had worked at Blue Origin, the rocket company founded by Amazon’s billionaire CEO Jeff Bezos. (Noone went on to work at Elon Musk’s SpaceX as well.) Both Blue Origin and SpaceX utilize 3-D printing—aka additive manufacturing—to construct some parts and tools for their launch vehicles, but presently still rely on more traditional techniques to make most parts of their rockets. Given that 3-D–printed components can be lighter, stronger and less complex than their more typically manufactured counterparts, Ellis and Noone saw and seized an opportunity. They decided to use the process to build what they hoped would be cheaper, better rockets to provide bargain-basement access to orbit, designing custom print heads and software to realize their disruptive dream.

After a stealthy period in which they raised upward of $40 million in venture capital and created a 20,000-square-foot facility in Los Angeles, Relativity Space is now printing and testing ever-larger rocket components. The company’s proprietary rocket engine boasts only two parts instead of the 2,600 it would contain if built via standard industry techniques. And just last month the company signed a $30-million, 20-year agreement with NASA Stennis Space Center for exclusive use of a 25-acre rocket-testing complex. Four test stands on the site could, in principle, allow Relativity Space to put enough engines through their paces to build 36 rockets per year—and the partnership also allows room for the company to expand its acreage at Stennis 10-fold. That growth may be necessary sooner rather than later, as other companies make ambitious plans to fill low Earth orbit with tens of thousands of new satellites for telecommunications, Earth observation and more.

Relativity Space has yet to build, let alone fly, an entirely 3-D–printed rocket, but Ellis is confident the company will launch its first one as early as 2020, followed shortly thereafter by commercial cargo flights. If all goes well, he says, this would be only the first step toward the company’s ultimate goal of making humanity multiplanetary, by printing and launching rockets from future settlements on Mars.

Scientific American spoke with Ellis about his plans for Relativity Space, the emerging market for midsize satellites in low Earth orbit and the company’s long-term vision for Mars-based manufacturing.

[An edited transcript of the interview follows.]

How is Relativity different from other rocket companies?

At its core, Relativity is creating an entirely new and fundamentally better process to build and fly rockets. We’re working to automate as much of the rocket design and production process as possible, by 3-D printing as much of a rocket as possible. We view 3-D printing as the future for all rocket production and aerospace manufacturing, because of how much it reduces the labor and part count of these very complex products. We’re developing a platform that combines software, machine learning, metallurgy and the largest metal 3-D printer in the world, which we call Stargate. Thanks to our proprietary printheads, software and metal-deposition process, Stargate can make on the order of 10-foot-diameter and 20-foot-tall parts, and it can do it 20 to 30 times faster than other more traditional 3-D printers. Right now it’s sized to the building we’re in at our headquarters in Los Angeles, but it’s designed to grow over time, both in hardware and software, to print bigger things even faster.

How fast can you print now?

We’re getting to the point where we can make a rocket structure—all the parts for a nearly 100-foot-tall, 7-foot-wide rocket—from scratch in 30 days. Once the parts are printed, the target is to assemble, test, integrate and fly the rocket within another 30 days. So we are en route to making an entire rocket from raw material to flight within 60 days. We’re using Stargate to make our own launch vehicle, Terran 1, as well as a rocket engine called Aeon 1. Those will be the basis of our launch service, which will initially carry satellites as heavy as 1,250 kilograms to low Earth orbit. We are planning to announce our launch site by the end of the year, and we expect our first full-scale, full-performance test flight to occur in late 2020, with commercial service beginning in early 2021.

How much would launches cost?

A launch on our rocket will be $10 million. That makes it highly cost-competitive with rockets that are much larger, such as Russia’s Soyuz, Europe’s Vega and India’s Polar Satellite Launch Vehicle. Those are the players we see ourselves up against in the near term. Some other start-ups are offering smaller launchers—like Rocket Labs’ Electron rocket—that can launch payloads more like 150 kilograms, but our rocket could be two to three times cheaper on just a per-satellite basis when we launch multiple payloads at once.

It seems like you would need to launch a lot of satellites to make the economics work. And most satellites are much heavier than 1,250 kilograms—or much lighter, like CubeSats and other microsatellites, right? Where is the demand for this?

Overall, we’re seeing more demand for bigger satellites versus smaller ones, as long as you can build and fly them cheaply. And we’re seeing satellites get larger as competition grows. Bigger optics, bigger sensors and bigger solar arrays offer disproportionately better performance. But we’re also seeing growing numbers of satellite “constellations” in low Earth orbit for telecommunications or Earth imaging, each consisting of dozens or hundreds or even thousands of midsized spacecraft. This is part of a big change in the space industry, a shift from very big satellites in geosynchronous orbits, which are almost 36,000 kilometers high, to constellations of midsized satellites which fly less than 1,000 kilometers high.

What is driving that demand, and how exactly does Relativity plan to fulfill it?

The push to reduce communications latency is a big part of it—having your delay go from hundreds to tens of milliseconds makes many applications competitive with terrestrial options in a way that just wasn’t possible before. For low Earth orbit, you need a very large number of satellites in multiple orbital planes to get frequent, full coverage—you really have to blanket the globe so you can collect and relay massive reams of data all over. SpaceX, for instance, plans to use its Falcon rockets to create Starlink, a constellation of thousands of satellites for high-speed worldwide broadband, but it’s not the only company with those sorts of plans. And constellations competing with SpaceX may not want to ride on a Falcon. So we’re going after two services—the ability to resupply constellations with new, midsize satellites as old ones fail, and the ability to deploy entire orbital planes full of smaller, 50- to 100-kilogram satellites in a single launch. We can do both very cost-effectively, and in fact already have more than a billion dollars in commitments to orders from a mix of commercial entities as well as governments. We can’t talk too much yet about those due to nondisclosure agreements, because the constellation market is so competitive. But we’ll be announcing our partners as things move forward.

Apparently, you are in this for the long haul. Does your 20-year lease at NASA’s Stennis Space Center mean you are already planning for the late 2030s?

Well, the way the agreement is technically structured is a 10-year agreement with a 10-year option, where the option is on us to continue or not. So that de facto ends up being 20 years because we’d love to stay and keep expanding! The whole point of our design and manufacturing process is that it can scale up or down quite easily. So we would use this site to just keep developing the most effective, best launch vehicle for wherever the market goes, assuming that streamlining that process through robotics and software automation is something that will be desired in the future. Ultimately, we want to use our approach to support a long-term human presence on Mars.

Why Mars? What do you see as Relativity’s role there?

So, this is the reason we started the company. You have Elon Musk and SpaceX saying, “Hey, we want to go to Mars to set up an entirely new society there.” And with NASA’s help they have had years of stunning results and progress. But I would say that no other company has come forward to express or support that same goal. And it’s our belief we’ll need dozens—if not hundreds—of companies to be inspired and to contribute to that long-term mission. Most of the activity right now seems to be focused on just figuring out how to get people to Mars, but someone needs to focus on how to leverage a very small labor force to build out an entire permanent society on another planet. That’s what Relativity is going to focus on: How to make something sustainable and scalable quickly, with low labor. Think about what sort of factory you could set up on Mars. You would want it to be lightweight with a small footprint, and for it to be flexible enough to make a wide range of products while adapting to unfamiliar conditions. What I have just described is a 3-D printer, and that’s what we want to build. Everything we’re doing with automating and building rockets and engines in our factory on Earth, we view as stepping-stones toward figuring out how to do this on Mars. Eventually, we’ll want to make rockets there to fly things back. Part of a society on Mars becoming self-sustaining will be the production and export of goods. When that happens, it will be a pivotal turning point in human history. And before then, of course, you could make other industrial products to help build up this whole society on another planet. Pursuing this has the nice side benefit of also creating a really kick-ass business on Earth.